Early GLP-1 Exposure Carries Different Pregnancy Signals by Indication
A Danish nationwide cohort of 756,636 singleton pregnancies has drawn a sharper line between GLP-1 receptor agonist exposure for diabetes and exposure for weight management. Early use of semaglutide or liraglutide around conception was linked to preterm birth only when the drugs had been prescribed for diabetes, not when they were being used for weight loss. The pattern shifts attention away from the medications alone and toward the metabolic disease that brought them into the pregnancy history in the first place.
This distinction matters because semaglutide, sold as Ozempic and Wegovy, and liraglutide, sold as Saxenda, have moved rapidly into routine use. Their appetite-suppressing and glucose-lowering effects have made them central to obesity and diabetes care, but pregnancy safety has remained unsettled. The current cohort does not prove harm from the drugs themselves. It does show that the underlying diagnosis cannot be ignored when obstetric risk is assessed.
A Nationwide Cohort That Captured Real-World Exposure
Investigators used Denmark’s health registries to identify pregnancies ending in delivery between October 1, 2009, and December 31, 2023. The cohort included 480,231 women and 756,636 pregnancies. Exposure was defined by prescription fills for semaglutide or liraglutide during the eight weeks before or after the first day of the last menstrual period, creating a 16-week window that brackets conception and the earliest phase of embryonic development.
Only 529 pregnancies met the exposure definition. Even with that small exposed group, the data set was large enough to support adjusted analyses across multiple obstetric outcomes, including pre-eclampsia, gestational diabetes, large-for-gestational-age birth, stillbirth, and placental complications. Maternal age, body mass index, smoking status, education, geographic location, calendar time, and pre-existing diabetes were all accounted for in the models.
One signal survived adjustment. The rest largely faded.
Preterm Birth Appeared Only in the Diabetes Group
The most clinically relevant finding centered on early delivery. Among women prescribed liraglutide or semaglutide for diabetes, the adjusted odds of premature birth rose sharply: 70% higher with liraglutide and 84% higher with semaglutide compared with unexposed pregnancies. The accompanying announcement also described estimated increases of about 11% for semaglutide and 9% for liraglutide among women with pre-existing diabetes. However those numbers are presented, the direction is the same: the elevated risk was confined to the diabetes subgroup.
Among women taking the same drugs for weight management, no significant association emerged for premature birth or for the other pregnancy complications examined. Early unadjusted comparisons had suggested broader obstetric risk, but those associations lost statistical significance after adjustment for confounders. The clean separation between treatment indication and outcome is the central finding of the study.
The biological interpretation is plausible. Diabetes is already associated with vascular dysfunction, placental abnormalities, fetal overgrowth, hypertensive disease, and spontaneous or indicated early delivery. In that setting, a medication history may be acting as a marker of a high-risk metabolic environment rather than as the direct cause of prematurity. By contrast, women taking GLP-1 therapy solely for weight control may have had fewer of the disease-related factors that push pregnancies toward early delivery.
Why the Indication Matters in Counseling
The study arrives at a moment when prescriptions for GLP-1 drugs have expanded quickly and inadvertent exposure in early pregnancy is becoming harder to avoid. Guidance currently recommends stopping treatment eight weeks before a planned pregnancy, but that advice is based on early animal work and limited human evidence. Real-world exposure data have been thin, which has left clinicians with more caution than clarity when a patient discovers she was taking semaglutide or liraglutide before recognizing she was pregnant.
For counseling, the message is more nuanced than reassurance or alarm. The absence of a detected risk signal in the weight-management group should not be read as proof of safety during pregnancy. These medications remain unestablished in pregnancy, and the cohort was observational rather than randomized. Still, the data suggest that a woman who used a GLP-1 receptor agonist for obesity treatment does not carry the same observed preterm birth signal seen in women whose pregnancies were complicated by diabetes.
That difference can shape the conversation in clinic. A patient with diabetes needs counseling that includes the baseline obstetric risk attached to glycemic disease itself, along with the possible contribution of medication exposure. A patient using the same drug for weight loss may need a different discussion, one focused on inadvertent early exposure and the absence of a clear adverse signal in this dataset.
Strengths, Limits, and What Remains Unsettled
The size of the study is its strongest feature. More than three-quarters of a million pregnancies were examined, giving this analysis far more reach than the small case series and anecdotal reports that have dominated the field so far. Registry-based data also reduce recall bias and allow researchers to study outcomes across a whole national population.
Several limits remain important. Prescription filling does not guarantee ingestion, even if the high cost of these injections makes adherence likely. Medication exposure was inferred from pharmacy records rather than confirmed biologically. The study also cannot separate the effect of diabetes from the effect of the drugs with complete certainty, even though the pattern strongly points toward the underlying disease as the main driver of risk.
The findings do not justify changing existing recommendations to discontinue GLP-1 therapy before a planned pregnancy. They do justify more careful stratification in future research. Women using these medications for diabetes should not be pooled with women using them for weight management if the goal is to understand reproductive safety. Merging those groups would blur a clinically meaningful difference and could mislead counseling.
That is the core lesson from the Danish cohort: the same drug can look different once the disease behind the prescription is separated out. For now, the safety question in pregnancy remains open, but the signal around preterm birth is no longer being interpreted in a vacuum.